The long term goal is to identify molecular mechanisms of cancer initiation and progression and hence to improve cancer prognosis and therapy. Smads are transcription factors initially identified as signaling molecules for the transforming growth factor ? (TGF?) superfamily. We found that Smad2 and Smad4 are frequently lost in human skin cancer and that keratinocyte-specific deletion of Smad2 or Smad4 in mice increases susceptibility to skin cancer. In contrast, Smad3 loss is uncommon in human skin cancers and mice with Smad3 ablation are resistant to skin cancer. The goal is to understand the mechanisms by which individual Smads affect carcinogenesis. Aim 1 is to assess whether loss of Smad2 or Smad4 increase genomic instability and promotes skin cancer formation and progression. We will induce skin tumor formation in our keratinocyte-specific Smad2 and Smad4 knockout mice and then delete Smad2 or Smad4 at defined stages of tumor development. Pathological and molecular alterations associated with genomic instability will be compared among stage-matched tumors from wild type, Smad2-/- and Smad4-/- mice. Transcriptional targets of Smad2 and Smad4 involved in genomic stability will also be identified. Aim 2 is to assess whether Smad3-dependent TGF? signaling contributes to tumor promotion in tumors with Smad4 loss. Keratinocyte-specific Smad4 knockout mice will be bred into a Smad3-/- background to abolish Smad3-dependent TGF? signaling in both tumor epithelia and stroma. We will then assess whether Smad3 deletion reduces TGF?1-mediated tumor promotion effect in Smad4-/- tumors. Smad3 target genes important in tumor promotion will also be identified. Aim 3 is to assess whether loss of Smad2 and Smad4 cooperate to promote skin carcinogenesis. We will generate mouse models in which both Smad2 and Smad4 will be ablated or knocked down. Mechanisms of cooperation between Smad2 and Smad4 loss will be identified by comparing pathological and molecular alterations in spontaneous and experimentally-induced tumors in these double knockout mice with those in a Smad2-/- or Smad4-/- background. The proposed studies will identify molecular targets of Smads in carcinogenesis, which will provide insights into cancer prognosis and future targeted therapies. The mouse models generated in this application will provide in vivo models that closely mimic naturally occurring human cancers, and thus can be used to test novel therapeutic approaches in the future.